Full Output Enforcement
leonxlnx
Not a design tool at all: it stops the model abbreviating. It counts what was asked for, writes every piece of it, and treats an elided middle, a stub comment or an offer to carry on if you want as a failed answer. Reach for it when truncation is the bug, whatever the subject happens to be.
Skills it gets confused with
These share a family with Full Output Enforcement, which is another way of saying they are the ones you might reach for by mistake.
- Academy GuideQuestions about how to use Claude well. Checks the official guidance before answering rather than improvising advice about the product.
- Claude APIBuilding something that calls Claude: model ids, pricing, streaming, tool use, prompt caching, token counting. Reach for it before guessing an API detail from memory.
- Discernment NudgeFires after advice is given, prompting the reader to check it rather than act on it blindly. Nothing a user asks for directly.
- Skill CreatorMaking skills, not using them: package a workflow you keep repeating into a skill, then test whether it actually fires when it should.
Knowing what Full Output Enforcement does is the easy half. Telling it apart from the others under time pressure is the game.
Today's sessionFull Output Enforcement is part of leonxlnx/taste-skill. Licence: MIT. The description above was written for this game, not taken from the skill.